Linear calculation unit, module, circuit and method based on memristor

Abstract

The invention belongs to the field of microelectronic devices and circuits, and particularly discloses a memristor-based linear calculation unit, module, circuit and method, the linear calculation unit comprises an f function circuit, a nonlinear memristor and a current-voltage conversion circuit which are cascaded; the f function circuit is used for converting a voltage signal of an input port of the circuit into a voltage signal of an output port of the circuit based on a f function, and the f function is an inverse function of a nonlinear function; the nonlinear memristor is used for converting a voltage signal of an input port of the device into a current signal of an output port of the device based on a nonlinear function; the current and voltage conversion circuit is used for converting a current signal of the input port of the circuit into a voltage signal of the output port of the circuit. According to the application, the linear relation between the input voltage and the output voltage of the linear calculation unit can be ensured, and linear analog calculation based on the nonlinear memristor is realized.

Description

Technical Field

[0001] The present application belongs to the field of microelectronic devices and circuits, and more specifically, to a linear computing unit, module, circuit and method based on a memristor. Background Art

[0002] In the post-Moore era, intelligent computing applications are rapidly increasing their demands for computing power and energy efficiency of computing systems. However, the traditional von Neumann computer with separated storage and computing is subject to data transmission bottlenecks, and performance improvement is severely restricted. The memristor-based storage-computing integrated architecture integrates computing and storage in the same device, becoming an emerging technology path to break through the bottleneck of the storage wall and realize high-efficiency, low-latency computing systems.

[0003] Currently, due to the influence of leakage current paths, memristors often need to rely on gating devices such as transistors or gating tubes to achieve large-scale integration. These integration methods not only increase the complexity of the preparation process, but also sacrifice the array integration density, resulting in a loss of system computing power and energy efficiency. Nonlinear memristors suppress leakage current based on interface engineering / barrier engineering, so high-density integration can be achieved without introducing gating devices, making them ideal devices for building storage and computing integrated systems.

[0004] However, since nonlinear memristors have a nonlinear IV relationship and do not follow Ohm's law, it is impossible to implement linear in-situ analog computing through linear analog input, and using binary input will cause serious performance loss. Therefore, in order to give full play to the high energy efficiency and high-density storage and computing advantages of nonlinear memristors, it is urgent to realize linear analog computing based on nonlinear memristors. Summary of the invention

[0005] In view of the defects of the prior art, the purpose of the present application is to realize linear analog calculation based on nonlinear memristor.

[0006] To achieve the above-mentioned object, in a first aspect, the present application provides a memristor-based linear computing unit, the linear computing unit comprising: a cascaded f-function circuit, a nonlinear memristor and a current-voltage conversion circuit;

[0007] The f function circuit is used to convert the voltage signal of the input port of the circuit into the voltage signal of the output port of the circuit based on the f function, and the f function is the inverse function of the nonlinear function;

[0008] The nonlinear memristor is used to convert a voltage signal at an input port of the device into a current signal at an output port of the device based on a nonlinear function;

[0009] The current-to-voltage conversion circuit is used to convert the current signal at the input port of the circuit into a voltage signal at the output port of the circuit.

[0010] It can be understood that a nonlinear function is used to fit the IV relationship of the nonlinear memristor, and an inverse function of the nonlinear function (also f function) is constructed. When performing analog calculations, the f function circuit can convert the voltage signal of the input port into a voltage signal of the output port based on the f function, and the nonlinear memristor can convert the voltage signal of the input port into a current signal of the output port based on the nonlinear function. The f function is the inverse function of the above nonlinear function, which can ensure that there is a linear relationship between the voltage of the input port of the f function circuit and the current of the output port of the nonlinear memristor. The current signal can then be converted into a voltage signal through the current-voltage conversion circuit, which can ensure that there is a linear relationship between the input voltage and the output voltage of the linear calculation unit, thereby realizing linear analog calculation based on the nonlinear memristor, breaking through the key performance bottleneck of linear calculation achieved by nonlinear devices, and realizing the simultaneous improvement of computing power and energy efficiency for intelligent computing applications in the post-Moore era.

[0011] In a possible implementation, the nonlinear function is an exponential function, and accordingly, the f function is a logarithmic function. The f function circuit includes: a cascaded logarithmic operation circuit and a first voltage amplification circuit;

[0012] The logarithmic operation circuit is used to convert the voltage signal of the input port of the circuit into the voltage signal of the output port of the circuit based on the logarithmic function;

[0013] The first voltage amplifier circuit is used to amplify the voltage signal of the input port of the circuit and output the amplified voltage signal through the output port of the circuit.

[0014] In a possible implementation, the logarithmic operation circuit includes: a first resistor, a diode and a first operational amplifier;

[0015] The first resistor is electrically connected to the non-inverting input terminal (Non-inverting Input, generally marked as "+") of the first operational amplifier, the inverting input terminal (Inverting Input, generally marked as "-") of the first operational amplifier is grounded, the anode of the diode is electrically connected to the non-inverting input terminal of the first operational amplifier, and the cathode of the diode is electrically connected to the output terminal of the first operational amplifier;

[0016] Accordingly, the first voltage amplifying circuit includes: a second resistor, a third resistor and a second operational amplifier;

[0017] The first end of the second resistor is electrically connected to the output end of the first operational amplifier, the second end of the second resistor is electrically connected to the non-inverting input end of the second operational amplifier, the inverting input end of the second operational amplifier is grounded, the first end of the third resistor is electrically connected to the non-inverting input end of the second operational amplifier, and the second end of the third resistor is electrically connected to the output end of the second operational amplifier.

[0018] In a possible implementation, the current-to-voltage conversion circuit includes: a cascaded transimpedance amplifier circuit and a second voltage amplifier circuit.

[0019] In a possible implementation, the transimpedance amplifier circuit includes: a fourth resistor and a third operational amplifier;

[0020] A first end of the fourth resistor is electrically connected to a non-inverting input terminal of the third operational amplifier, a second end of the fourth resistor is electrically connected to an output terminal of the third operational amplifier, and an inverting input terminal of the third operational amplifier is grounded;

[0021] Accordingly, the second voltage amplifying circuit includes: a fifth resistor, a sixth resistor and a fourth operational amplifier;

[0022] The first end of the fifth resistor is electrically connected to the output terminal of the third operational amplifier, the second end of the fifth resistor is electrically connected to the non-inverting input terminal of the fourth operational amplifier, the inverting input terminal of the fourth operational amplifier is grounded, the first end of the sixth resistor is electrically connected to the non-inverting input terminal of the fourth operational amplifier, and the second end of the sixth resistor is electrically connected to the output terminal of the fourth operational amplifier.

[0023] In a second aspect, the present application provides a memristor-based linear computing module, the linear computing module comprising: m f function circuits, k current-voltage conversion circuits and a nonlinear memristor array, the nonlinear memristor array being composed of m rows × k columns of nonlinear memristors;

[0024] The input ports of the nonlinear memristors in the same row are connected to the output ports of the same f-function circuit;

[0025] The output ports of the nonlinear memristors in the same column are connected to the input ports of the same current-to-voltage conversion circuit;

[0026] The f function circuit is used to convert the voltage signal of the input port of the circuit into the voltage signal of the output port of the circuit based on the f function, and the f function is the inverse function of the nonlinear function;

[0027] The nonlinear memristor is used to convert a voltage signal at an input port of the device into a current signal at an output port of the device based on a nonlinear function;

[0028] The current-to-voltage conversion circuit is used to convert the current signal at the input port of the circuit into a voltage signal at the output port of the circuit.

[0029] In a third aspect, the present application provides a memristor-based linear computing circuit, the linear computing circuit comprising: a main controller, an input circuit, an output circuit, and a linear computing module described in the second aspect or any possible implementation of the second aspect;

[0030] The main controller is used for inputting the data to be calculated into the linear calculation module through the input circuit; and receiving the calculation result of the linear calculation module through the output circuit.

[0031] In a possible implementation, the input circuit includes: a cascaded input buffer module, a data encoding module, and a digital-to-analog conversion module;

[0032] The data encoding module is used to match the input data of this module with the voltage operating range of the nonlinear memristor.

[0033] In a possible implementation, the output circuit includes: a cascaded analog-to-digital conversion module and an output buffer module.

[0034] In a fourth aspect, the present application provides a linear calculation method based on a memristor, the method being applied to the linear calculation circuit described in the third aspect or any possible implementation manner of the third aspect, the method comprising:

[0035] Inputting the data to be calculated into the linear calculation module through the input circuit;

[0036] The calculation result of the linear calculation module is received through the output circuit.

[0037] It can be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0038] In general, the above technical solutions conceived by this application have the following beneficial effects compared with the prior art:

[0039] (1) The present application provides an analog storage-computing integrated circuit, system and operation method thereof, wherein an f-function circuit and a g-function circuit are constructed by fitting the nonlinear IV characteristics of the device, and the f-function circuit converts the linear voltage input to the nonlinear memristor into a nonlinear voltage input. The linear current output is obtained by the IV characteristics of the device, and finally the current output is converted into a voltage and amplified by the g-function circuit. This design can realize linearized analog computing based on nonlinear memristors, give full play to the performance advantages of nonlinear devices, and solve the defects of existing storage-computing integrated units in terms of integration density and computing energy efficiency.

[0040] (2) The computing unit architecture and operation method provided by the present application have the characteristics of generalized application, that is, any nonlinear memristor based on different physical mechanisms and following different IV relationships can realize linear analog calculation under the computing unit architecture and calculation method designed by the present application by fitting and analyzing the IV relationship, constructing a mathematical model, and designing the f function circuit and the g function circuit. However, the f function circuit and the g function circuit structure or circuit parameters are different from those of the present application. Therefore, the present application can be extended to realize an efficient analog computing system for any nonlinear memristor. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a schematic diagram of the structure of a nonlinear memristor provided in an embodiment of the present application;

[0042] Figure 2 is a schematic diagram of the device resistance state conversion characteristics of the nonlinear memristor provided in an embodiment of the present application;

[0043] Figure 3 Schematic diagram of the multi-value programming characteristics and nonlinear IV relationship of the nonlinear memristor provided in the embodiment of the present application;

[0044] Figure 4 It is a schematic diagram of a circuit structure of a linear computing unit implemented based on a nonlinear memristor provided in an embodiment of the present application;

[0045] Figure 5 is a schematic diagram of the f function circuit structure and circuit characteristics provided in an embodiment of the present application;

[0046] Figure 6 is a schematic diagram of the g-function circuit structure and circuit characteristics provided in an embodiment of the present application;

[0047] Figure 7 is a schematic diagram of an array expansion result obtained based on a linear calculation unit and its input-output relationship provided in an embodiment of the present application;

[0048] Figure 8 is a schematic diagram of the structure of a linear analog calculation circuit provided in an embodiment of the present application;

[0049] Fig. 9 It is a schematic diagram of the process of the linear analog calculation circuit provided in the embodiment of the present application completing the linear analog calculation. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0051] The terms "first" and "second" in the specification and claims herein are used to distinguish different objects rather than to describe a specific order of the objects. For example, a first voltage amplifier circuit and a second voltage amplifier circuit are used to distinguish different voltage amplifier circuits rather than to describe a specific order of the voltage amplifier circuits.

[0052] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0053] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc.; multiple elements refer to two or more elements, etc.

[0054] The embodiments of the present application are described below in conjunction with the drawings in the embodiments of the present application.

[0055] Example 1

[0056] Figure 1 The basic structure of a nonlinear memristor is demonstrated, and the device consists of an upper electrode, a barrier layer, a resistive layer and a lower electrode. Figure 2 The resistance state transition characteristics of the device are shown, and the scanning direction of the curve is indicated, where SET represents the device changing from a high resistance state to a low resistance state (set to "1"), and RESET represents the device changing from a low resistance state to a high resistance state (set to "0"). The current suppression characteristics of the device during the reverse (negative voltage) scanning process can be observed in the resistance change curve. This allows the device to suppress the leakage current path in the array when the array is integrated, so the device can achieve high-density integration without using a gating device (such as a transistor or a gating tube).

[0057] Due to the influence of the device mechanism, the device does not have a linear IV characteristic, such as Figure 3 When the device is placed in G 1 , G 2 …G mWhen the device changes to different conductance (resistance) states, a linearly increasing voltage input is applied to the device, and the device exhibits a nonlinear current response, that is, the IV relationship of the device does not satisfy Ohm's law. This characteristic makes it impossible for nonlinear memristors to perform linear analog calculations based on Ohm's law and Kirchhoff's current law, and linear analog calculations are the basis for memristor arrays to accelerate intelligent computing applications such as neural networks. Therefore, nonlinear memristors are restricted in achieving high-efficiency storage and computing hardware. However, the different conductance states of nonlinear memristors follow a unified nonlinear function fitting relationship within a certain voltage range. For example, the nonlinear function relationship is I W =a·G·exp(b·V W ), where V W is the input voltage of the device, I W This makes it possible to construct a corresponding inverse function circuit (ie, the f function circuit below) based on the fitted function relationship to support the nonlinear device to implement linear analog calculation, as described in Example 2.

[0058] Example 2

[0059] Figure 4 A linear computing unit based on a nonlinear memristor is presented, the purpose of which is to apply a linear input voltage to the nonlinear memristor to obtain a linear output voltage response. The linear computing unit consists of three main parts: an f function circuit, a nonlinear memristor, and a g function circuit (that is, the above-mentioned current-voltage conversion circuit). The input port of the f function circuit applies the input voltage V X The output port is connected to the input port (i.e., the upper electrode) of the nonlinear memristor. The input port of the g function circuit is connected to the output port (i.e., the lower electrode) of the nonlinear memristor, and the output port obtains the output result V of the calculation unit. y Through circuit design and parameter adjustment, the input-output relationship of the f function circuit is Where V X is the circuit input voltage, V W is the circuit output voltage. The input and output relationship of the g function circuit is Among them I W is the circuit input current, V y is the circuit output voltage. Combined with the nonlinear IV relationship of the device, the input-output relationship of the linear calculation unit can be obtained as V y =G·V X , where G is the current conductivity state of the device. Therefore, the input-output response of the linear calculation unit is a linear relationship related to the conductivity state of the device.

[0060] The f function circuit structure designed in Example 2 is as follows Figure 5As shown in the figure, the circuit is composed of two stages of operational amplifier circuits. The first stage circuit uses resistor R1, diode D1 and operational amplifier T1 to form a logarithmic function calculation circuit. The output voltage V O With input voltage V X The relationship between V O =-U T ·ln(V X / I S ·R 1 ) Among them, U T is the temperature constant of D1, which is 0.026mV at room temperature, I S is the reverse saturation current of D1. Preferably, I S ·R 1 =1V, that is, V O =-U T ·ln(V X ). The second stage circuit is a reverse amplifier circuit composed of an operational amplifier T2, resistors R2 and R3. Optionally, the amplification factor of the reverse amplifier circuit is R2 / R3=1 / b·U T , then the input-output relationship of the f function circuit can satisfy

[0061] The g function circuit designed in Example 2 is as follows: Figure 6 As shown in FIG. 1 , the circuit is also composed of two stages of operational amplifier circuits. The first stage circuit is composed of an operational amplifier T3 and a resistor R4 to form a transimpedance amplifier, which is used to convert the output current of the nonlinear memristor into a voltage and amplify it. Therefore, the amplification factor of the transimpedance amplifier is R4 = 10mΩ, and the specific value needs to be selected according to the device conductance value. The second stage circuit is an inverting amplifier composed of an operational amplifier T4, resistors R5 and R6, and the amplification factor is R6 / R5 = 1 / a.

[0062] Further, Example 2 is based on Figure 4 The linear computing unit shown in FIG. 1 is expanded in array to form a linear (analog) computing module, such as Figure 7 As shown. Among them, the nonlinear memristor is integrated into a device array based on a cross structure. The array size is m rows × k columns, the row lines are the input ports of the array, and the column lines are the output ports of the array. In order to realize parallel analog calculation, each row line of the array is connected to an f function circuit, and each column line of the array is connected to a g function circuit. According to the input-output characteristics of the above linear calculation unit, assuming that the input voltage is [V1, V2, V3, …, Vm], assuming that the device conductivity value of a certain column is [G1, G2, G3, …, Gm], then the output voltage corresponding to the column is Therefore, the linear computing module implements linear matrix / vector calculations and can be used to accelerate intelligent computing applications such as neural networks.

[0063] Example 3

[0064] Figure 8 A linear (analog) computing circuit is presented, which includes a main controller, an input cache module, a data encoding module, a digital-to-analog conversion module, an analog-to-digital conversion module, and an output cache module in addition to a linear computing module (the module includes an f function circuit, a nonlinear memristor array, and a g function circuit). Specifically, the input cache module is used to store data input to the linear computing module; the output end of the input cache module is connected to the input end of the data encoding module, and the data encoding module is used to match the input data and the voltage operating range of the nonlinear memristor; the output end of the data encoding module is connected to the input end of the digital-to-analog conversion module, and the digital-to-analog conversion module maps the encoded data to voltage when the computing unit is running, and the output end of the digital-to-analog conversion module is connected to the input end of the f function circuit in the linear computing module. At the same time, the input end of the analog-to-digital conversion module is connected to the output end of the g function circuit in the linear computing module, and is used to convert the calculation result (voltage quantity) of the linear computing module into a digital quantity; the output end of the analog-to-digital conversion module is connected to the input end of the output cache module, and the output cache module is used to store the calculation result; it should be noted that the operation of the input cache module and the output cache module are both controlled by the main controller.

[0065] Based on the designed circuit architecture, Embodiment 3 further provides an operation method of the calculation unit, including the following steps S1 to S4.

[0066] Step S1: The main controller controls the input buffer module to load input data.

[0067] Step S2: The input data is matched with the voltage operating range of the nonlinear memristor via the data encoding module, and then converted into the voltage input of the linear calculation module (the module includes the f function circuit, the nonlinear memristor array and the g function circuit) through the digital-to-analog conversion module.

[0068] Step S3: The linear input voltage is converted into a nonlinear voltage via the f function circuit and input into the nonlinear memristor array. The in-situ analog calculation junction (current) of the nonlinear memristor array is output to the g function circuit, and a linear output voltage is obtained after conversion and amplification.

[0069] Step S4: The linear output voltage is converted into a digital quantity after analog-to-digital conversion and stored in the output buffer module. The calculation is completed, and the main controller can receive the calculation result of the linear calculation module through the output buffer module.

[0070] Fig. 9 The analog quantity change relationship between the operation steps described in Example 3 is shown. After S1-S2, the input voltage V of the matching device nonlinear fitting interval is obtained by the digital-analog conversion unit.in , V in In step S3, the nonlinear memristor array input V is obtained through the f function circuit. W , V W After in-situ calculation by the nonlinear memristor array, the linear output current I is obtained W , current I W After being amplified by the g function circuit, the output voltage V OUT , V OUT In step S4, a digital output is obtained through analog-to-digital conversion and stored in an output buffer module.

[0071] It is understandable that during calculation, the input data is converted into a linear voltage input via the input cache module-data encoding module-digital-analog conversion module, and the voltage input is converted into a nonlinear voltage input via the f function circuit and the nonlinear memristor array is multiplied and accumulated. The analog calculation output of the nonlinear memristor array is amplified by the g function circuit to obtain a linear output voltage. The output voltage is finally stored in the output cache module via the analog-to-digital conversion circuit. The present application solves the technical defect that the nonlinear memristor array cannot realize linear analog calculation, and can fully develop the advantages of the nonlinear memristor array in terms of area efficiency and computing energy efficiency in practical applications.

[0072] It should be noted that the architecture of the linear (analog) computing circuit designed in Example 3 and the corresponding operation method have generalized application characteristics, that is, they are not only for the nonlinear memristor provided in Example 1. Any nonlinear memristor based on different physical mechanisms and following different IV relationships can be fitted and analyzed for the IV relationship, a mathematical model is constructed, and the f-function circuit and the g-function circuit are designed to realize linear analog computing under the architecture and calculation method provided in Example 3, except that the f-function circuit and the g-function circuit structure or circuit parameters are different from those in Example 3. Therefore, Example 3 can be extended to realize an efficient analog computing system for any nonlinear memristor.

[0073] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A linear computing unit based on a memristor, characterized in that: include: Cascaded f-function circuits, nonlinear memristors, and current-to-voltage conversion circuits; The f function circuit is used to convert the voltage signal of the input port of the circuit into the voltage signal of the output port of the circuit based on the f function, and the f function is the inverse function of the nonlinear function; The nonlinear memristor is used to convert a voltage signal at an input port of the device into a current signal at an output port of the device based on a nonlinear function; The current-to-voltage conversion circuit is used to convert the current signal at the input port of the circuit into a voltage signal at the output port of the circuit.

2. The memristor-based linear computing unit according to claim 1, characterized in that: The nonlinear function is an exponential function, and accordingly, the f function is a logarithmic function, and the f function circuit includes: a cascaded logarithmic operation circuit and a first voltage amplification circuit; The logarithmic operation circuit is used to convert the voltage signal of the input port of the circuit into the voltage signal of the output port of the circuit based on the logarithmic function; The first voltage amplifier circuit is used to amplify the voltage signal of the input port of the circuit and output the amplified voltage signal through the output port of the circuit.

3. The memristor-based linear computing unit according to claim 2, characterized in that: The logarithmic operation circuit comprises: a first resistor, a diode and a first operational amplifier; The first resistor is electrically connected to the non-inverting input terminal of the first operational amplifier, the inverting input terminal of the first operational amplifier is grounded, the anode of the diode is electrically connected to the non-inverting input terminal of the first operational amplifier, and the cathode of the diode is electrically connected to the output terminal of the first operational amplifier; Accordingly, the first voltage amplifying circuit includes: a second resistor, a third resistor and a second operational amplifier; The first end of the second resistor is electrically connected to the output end of the first operational amplifier, the second end of the second resistor is electrically connected to the non-inverting input end of the second operational amplifier, the inverting input end of the second operational amplifier is grounded, the first end of the third resistor is electrically connected to the non-inverting input end of the second operational amplifier, and the second end of the third resistor is electrically connected to the output end of the second operational amplifier.

4. The memristor-based linear computing unit according to claim 1, characterized in that: The current-to-voltage conversion circuit includes: a cascaded transimpedance amplifier circuit and a second voltage amplifier circuit.

5. The memristor-based linear computing unit according to claim 4, characterized in that: The transimpedance amplifier circuit comprises: a fourth resistor and a third operational amplifier; The first end of the fourth resistor is electrically connected to the non-inverting input terminal of the third operational amplifier, the second end of the fourth resistor is electrically connected to the output terminal of the third operational amplifier, and the inverting input terminal of the third operational amplifier is grounded; Accordingly, the second voltage amplifying circuit includes: a fifth resistor, a sixth resistor and a fourth operational amplifier; The first end of the fifth resistor is electrically connected to the output end of the third operational amplifier, the second end of the fifth resistor is electrically connected to the non-inverting input end of the fourth operational amplifier, the inverting input end of the fourth operational amplifier is grounded, the first end of the sixth resistor is electrically connected to the non-inverting input end of the fourth operational amplifier, and the second end of the sixth resistor is electrically connected to the output end of the fourth operational amplifier.

6. A linear computing module based on a memristor, characterized in that: include: m f function circuits, k current-voltage conversion circuits and a nonlinear memristor array, wherein the nonlinear memristor array is composed of m rows×k columns of nonlinear memristors; The input ports of the nonlinear memristors in the same row are connected to the output ports of the same f-function circuit; The output ports of the nonlinear memristors in the same column are connected to the input ports of the same current-to-voltage conversion circuit; The f function circuit is used to convert the voltage signal of the input port of the circuit into the voltage signal of the output port of the circuit based on the f function, and the f function is the inverse function of the nonlinear function; The nonlinear memristor is used to convert a voltage signal at an input port of the device into a current signal at an output port of the device based on a nonlinear function; The current-to-voltage conversion circuit is used to convert the current signal at the input port of the circuit into a voltage signal at the output port of the circuit.

7. A linear computing circuit based on a memristor, characterized in that: include: A main controller, an input circuit, an output circuit and a memristor-based linear computing module as claimed in claim 6; The main controller is used to input the data to be calculated into the linear calculation module through the input circuit; And receiving the calculation result of the linear calculation module through the output circuit.

8. The memristor-based linear computing circuit according to claim 7, characterized in that: The input circuit comprises: a cascaded input buffer module, a data encoding module and a digital-to-analog conversion module; The data encoding module is used to match the input data of this module with the voltage operating range of the nonlinear memristor.

9. The memristor-based linear computing circuit according to claim 7, characterized in that: The output circuit comprises: a cascaded analog-to-digital conversion module and an output buffer module.

10. A linear calculation method based on memristor, characterized in that: Applied to the memristor-based linear computing circuit as claimed in any one of claims 7 to 9, the method comprising: Inputting the data to be calculated into the linear calculation module through the input circuit; The calculation result of the linear calculation module is received through the output circuit.

Images